Microchannel plates are, for example, an essential component for fabrication of wafer tube image intensifiers. FIGS. 1-4 illustrate standard prior art devices and their operation. As shown in FIG. 1 a proximity-focussed wafer tube image intensifier 10 includes an input window 12 of glass or a fiber optic face plate onto the back of which is applied a photocathode 14. The microchannel plate 16 is spaced from and mounted parallel with the photocathode 14, and down stream of the microchannel plate 16 a phosphor screen 20 is provided on an output window 18 in the form of another fiber optics faceplate or glass. The input window 12 and output window 18 are mounted on opposite ends of a vacuum housing 22 with the microchannel plate 16 contained therebetween within the vacuum housing. The tube is provided with electrical leads for applying appropriate desired voltages to the photocathode 14, an input electrode 24 (see FIG. 2) on the front and an output electrode 26 (see FIG. 2) on the back of the microchannel plate 16 and phosphor screen 20.
The three main components of a wafer tube 10 are the photocathode 14, the microchannel plate 16, and the output phosphor screen 20. The photocathode 14 converts incident photons into photoelectrons. Generation-II wafer tubes use an alkali antimonide, positive affinity, photocathode. Generation-III wafer tubes use a GaAs, negative electron affinity, photocathode. The microchannel plate 16 serves as a high resolution electron multiplier which amplifies the photoelectron image. As used in an image intensifier the MCP typically has an electron gain of 100-1000. The amplified signal is accelerated by a 6 kv bias into the phosphor screen 20 which converts the electron energy into output light allowing the image to be viewed.
The microchannel plate 16 as shown enlarged in FIG. 2 consists of an array of miniature channel multipliers 28 of hollow glass fibers fused together and surrounded by a solid, glass border ring 30. As shown in FIG. 3 each channel multiplier 28 detects and amplifies incident radiation and particles such as electrons or ions. The channel multiplier concept is based on the continuous dynode electron multiplier first suggested by P. T. Farnsworth, U.S. Pat. No. 1,969,399. The channel multiplier 28 consists of a hollow tube coated on the interior surface by a secondary electron emitting semiconductor layer 32. This layer 32 emits secondary electrons in response to bombardment by electromagnetic radiation or particles such as electrons. The input and output metal electrodes 24 and 26 are provided on each end of the tube 28 to allow a bias voltage to be applied across the channel. This bias voltage creates an axial electric field which accelerates the emitted secondary electrons down the channel 28. The secondary electrons strike the wall again releasing additional secondary electrons. This process repeats as the electrons are accelerated down the channel. This results in amplification of the input photon or particle. A large pulse of electrons is emitted from the output end of the channel 28 in response to the input photon or particle.
In the typical microchannel plate 16, channel diameters can be as small as a few microns. For image intensification devices channel diameters are typically 10-12 microns. The channels typically have a length to diameter ratio of 40. The channel axes are typically biased at a small angle (5.degree.) relative to the normal to the MCP surface. The bias angle ensures that ions generated at the tube anode cannot be accelerated down the channel, but strike the channel wall near the back of the MCP. This reduces ion feedback noise in the MCP and eliminates ion feedback from the phosphor screen to the photocathode.
A typical plate may contain an active region 18 mm in diameter and contains over a million channels. The plate is fabricated from a glass wafer. The wafer is cut from a boule formed by fusing together glass fibers. The glass fibers are composed of a core glass surrounded by a clad glass of a different composition. After the glass wafers are sliced from the boule, the core glass is removed by a selective etching process thus forming the hollow channels. The plates are fired in hydrogen which reduces the exposed glass surface thereby forming a semiconducting layer on the channel wall surface. The thin silica layer 32 resides on the semiconducting layer forming the secondary electron emissive surface.
Traditionally, the input and output electrodes 24 and 26 are formed on each surface of the plate by deposition of a thin metallization layer. The layers thickness is typically on the order of 800 .ANG. for the input electrode 24 and 1100 .ANG. for the output electrode 26. FIG. 4 is an electron microscopic view of a cross sectioned MCP in the region of the output electrode. The metallization thickness (1100 .ANG.) is so thin relative to the channel diameter (10 microns) as to not be visible in the photograph. Nichrome or inconel are the commonly used electrode materials. These materials are used because of their good adhesion to the glass surface of the MCP.
The input electrode 24 is deposited by vacuum evaporation with a collimated beam of metal atoms. The beam is incident at a steep angle relative to the MCP surface to minimize penetration of the metal down the MCP channels. The MCP is rotated during the metallization process to result in uniform coverage of the plate surface and penetration of the channel. The practical limit is one half of a channel diameter penetration of the metal down the channel. It is desirable to limit the channel penetration as the commonly used metals, inconel or nichrome, have a very low secondary electron emission coefficient. If the primary particle or photon strikes the metallized channel wall a secondary electron may not be generated. Thus the gain of the MCP is lowered. More importantly the noise performance of the MCP suffers as some of the primary particles are not detected if they strike the metallized channel wall. The noise performance of the MCP is also degraded by the broad single particle gain distribution which results from the variation in gain depending upon whether the primary particle strikes the input metallization 24 or the secondary electron emitting layer 32.
The output electrode 26 is also deposited by vacuum evaporation with a collimated beam of metal atoms. In this case the incident angle is adjusted along with the MCP rotation to allow deeper penetration of the channel by the metal. Typically the metal penetrates 1.5 to 3.0 channel diameters. This is known as endspoiling to those familiar in the art of MCP manufacture. The gain of the MCP is reduced by this procedure. However this gain reduction is more than offset by other, desirable, characteristics which result from this procedure for MCPs which are used in image intensifiers. In particular, the output electron energy distribution of endspoiled MCPs is much more uniform than from plates with no endspoiling as described by N. Koshida "Effects of Electrode Structure on Output Electron Energy Distribution of Microchannel Plates", Rev. Sci. Instrum., 57(3), 354 (1986). This allows image intensifiers with higher resolution to be manufactured with end spoiled MCPs due to the improved electron optics which result from the uniform output electron energy distribution.
The improved emitted electron energy distribution which results from endspoiling is due to the fact that the majority of the emitted electrons are secondaries from the metallized channel walls which form the endspoiled region. These secondaries are given off when an electron emitted from farther up the channel is accelerated down the channel by the axial electric field and strikes the metallized region at the output of the channel. The axial electric field in the endspoiled region is zero due to the high conductivity of the metal. Therefore the emitted electrons are not accelerated after emission resulting in a more uniform emitted electron energy distribution.
The noise performance of an image intensifier is critical to its usefulness as a low light level imager. The noise performance is typically characterized by the noise factor, K.sub.f, of the image intensifier. The noise factor of an image intensifier has been considered to be largely determined by the noise performance of the MCP in the past. The noise factor can be defined by the following equation. ##EQU1## SNR is the signal-to-noise power ratio. SNR.sub.in is the SNR of the input electron flux to the MCP. In an image intensifier this is also the SNR of the photoelectron flux from the photocathode. SNR.sub.out is the SNR of the output photon flux from the image intensifier phosphor screen. Both ratios are measured over the same noise bandwidth. The noise factor can also be defined where SNR.sub.out is the SNR of the output electron flux from the MCP. In this instance the noise factor is that of the MCP alone. The noise factor results presented in this disclosure are given in terms of that for an image intensifier where SNR.sub.in is for the photoelectron flux from the photocathode and SNR.sub.out is for the photon flux from the intensifier phosphor screen.
The noise performance of a MCP based image intensifier can be further degraded by various feedback mechanisms. The feedback mechanisms which generate noise that have been considered in the past relate to internally generated ion feedback in the MCP or optical photon feedback from the phosphor screen as described by R. L. Bell "Noise Figure of the MCP Image Intensifier Tube", IEEE Trans. Elec. Dev. ED-22, No. 10, pages 821-829, October (1975). These ions can generate noise pulses when accelerated back toward the MCP input where secondary electrons are generated when the ions strike the channel wall. In the case of a Gen-II image intensifier the ions may be accelerated to the photocathode generating secondary electrons. In the Gen-III technology ion feedback from the MCP to the photocathode has been eliminated by applying a thin (50-100 .ANG.) film over the MCP input as described by H. K. Pollehn, "Image Intensifiers" Applied Optics and Optical Engineering, Vol. VI, 399, Academic Press, (1980). This film is semi-transparent to the photoelectrons, but will stop ions from bombarding the photocathode.
Optical photon feedback is avoided in a prior art image intensifier by ensuring that the aluminum metallization layer, which forms the anode of the tube and coats the phosphor, is sufficiently thick to completely stop penetration of light generated by the phosphor screen. This technique is effective and generally eliminates any significant feedback by optical photons to the MCP or photocathode. Optical photons, because of their low energy (2-3 eV), can also generate no more than one photoelectron upon impact with the MCP input or photocathode and thus cannot cause the large scintillations observed in an image intensifier. Phosphor screen to MCP wall ion feedback is somewhat limited in the prior art via the 5.degree. bias angle used by prior art MCPs.